[MATH-777] added NPointCrossover policy, thanks to Reid Hochstedler.

git-svn-id: https://svn.apache.org/repos/asf/commons/proper/math/trunk@1344030 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Thomas Neidhart 2012-05-29 22:24:24 +00:00
parent 129e53d466
commit 86e1cc0335
2 changed files with 291 additions and 0 deletions

View File

@ -0,0 +1,178 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.commons.math3.genetics;
import java.util.ArrayList;
import java.util.List;
import org.apache.commons.math3.exception.DimensionMismatchException;
import org.apache.commons.math3.exception.MathIllegalArgumentException;
import org.apache.commons.math3.exception.NotStrictlyPositiveException;
import org.apache.commons.math3.exception.NumberIsTooLargeException;
import org.apache.commons.math3.exception.util.LocalizedFormats;
import org.apache.commons.math3.random.RandomGenerator;
/**
* N-point crossover policy. For each iteration a random crossover point is
* selected and the first part from each parent is copied to the corresponding
* child, and the second parts are copied crosswise.
*
* Example (2-point crossover):
* <pre>
* -C- denotes a crossover point
* -C- -C- -C- -C-
* p1 = (1 0 | 1 0 0 1 | 0 1 1) X p2 = (0 1 | 1 0 1 0 | 1 1 1)
* \----/ \-------/ \-----/ \----/ \--------/ \-----/
* || (*) || || (**) ||
* VV (**) VV VV (*) VV
* /----\ /--------\ /-----\ /----\ /--------\ /-----\
* c1 = (1 0 | 1 0 1 0 | 0 1 1) X c2 = (0 1 | 1 0 0 1 | 0 1 1)
* </pre>
*
* This policy works only on {@link AbstractListChromosome}, and therefore it
* is parameterized by T. Moreover, the chromosomes must have same lengths.
*
* @param <T> generic type of the {@link AbstractListChromosome}s for crossover
* @since 3.1
* @version $Id$
*/
public class NPointCrossover<T> implements CrossoverPolicy {
/** The number of crossover points. */
private final int crossoverPoints;
/**
* Creates a new {@link NPointCrossover} policy using the given number of points.
*
* <p><b>Note</b>: the number of crossover points must be &lt; <code>chromosome length - 1</code>.
* This condition can only be checked at runtime, as the chromosome length is not known in advance.
* </p>
*
* @param crossoverPoints the number of crossover points
* @throws NotStrictlyPositiveException if the number of {@code crossoverPoints} is not
* strictly positive
*/
public NPointCrossover(final int crossoverPoints) {
if (crossoverPoints <= 0) {
throw new NotStrictlyPositiveException(crossoverPoints);
}
this.crossoverPoints = crossoverPoints;
}
/**
* Returns the number of crossover points used by this {@link CrossoverPolicy}.
*
* @return the number of crossover points
*/
public int getCrossoverPoints() {
return crossoverPoints;
}
/**
* Performs a N-point crossover. N random crossover points are selected and are used
* to divide the parent chromosomes into segments. The segments are copied in alternate
* order from the two parents to the corresponding child chromosomes.
*
* Example (2-point crossover):
* <pre>
* -C- denotes a crossover point
* -C- -C- -C- -C-
* p1 = (1 0 | 1 0 0 1 | 0 1 1) X p2 = (0 1 | 1 0 1 0 | 1 1 1)
* \----/ \-------/ \-----/ \----/ \--------/ \-----/
* || (*) || || (**) ||
* VV (**) VV VV (*) VV
* /----\ /--------\ /-----\ /----\ /--------\ /-----\
* c1 = (1 0 | 1 0 1 0 | 0 1 1) X c2 = (0 1 | 1 0 0 1 | 0 1 1)
* </pre>
*
* @param first first parent (p1)
* @param second second parent (p2)
* @return pair of two children (c1,c2)
* @throws MathIllegalArgumentException iff one of the chromosomes is
* not an instance of {@link AbstractListChromosome}
* @throws DimensionMismatchException if the length of the two chromosomes is different
*/
@SuppressWarnings("unchecked") // OK because of instanceof checks
public ChromosomePair crossover(final Chromosome first, final Chromosome second) {
if (!(first instanceof AbstractListChromosome<?> && second instanceof AbstractListChromosome<?>)) {
throw new MathIllegalArgumentException(LocalizedFormats.INVALID_FIXED_LENGTH_CHROMOSOME);
}
return mate((AbstractListChromosome<T>) first, (AbstractListChromosome<T>) second);
}
/**
* Helper for {@link #crossover(Chromosome, Chromosome)}. Performs the actual crossover.
*
* @param first the first chromosome
* @param second the second chromosome
* @return the pair of new chromosomes that resulted from the crossover
* @throws DimensionMismatchException if the length of the two chromosomes is different
* @throws NumberIsTooLargeException if the number of crossoverPoints is too large for the
* actual chromosomes
*/
private ChromosomePair mate(final AbstractListChromosome<T> first,
final AbstractListChromosome<T> second) {
final int length = first.getLength();
if (length != second.getLength()) {
throw new DimensionMismatchException(second.getLength(), length);
}
if (crossoverPoints >= length) {
throw new NumberIsTooLargeException(crossoverPoints, length, false);
}
// array representations of the parents
final List<T> parent1Rep = first.getRepresentation();
final List<T> parent2Rep = second.getRepresentation();
// and of the children
final ArrayList<T> child1Rep = new ArrayList<T>(first.getLength());
final ArrayList<T> child2Rep = new ArrayList<T>(second.getLength());
final RandomGenerator random = GeneticAlgorithm.getRandomGenerator();
ArrayList<T> c1 = child1Rep;
ArrayList<T> c2 = child2Rep;
int remainingPoints = crossoverPoints;
int lastIndex = 0;
for (int i = 0; i < crossoverPoints; i++, remainingPoints--) {
// select the next crossover point at random
final int crossoverIndex = 1 + lastIndex + random.nextInt(length - lastIndex - remainingPoints);
// copy the current segment
for (int j = lastIndex; j < crossoverIndex; j++) {
c1.add(parent1Rep.get(j));
c2.add(parent2Rep.get(j));
}
// swap the children for the next segment
ArrayList<T> tmp = c1;
c1 = c2;
c2 = tmp;
lastIndex = crossoverIndex;
}
// copy the last segment
for (int j = lastIndex; j < length; j++) {
c1.add(parent1Rep.get(j));
c2.add(parent2Rep.get(j));
}
return new ChromosomePair(first.newFixedLengthChromosome(child1Rep),
second.newFixedLengthChromosome(child2Rep));
}
}

View File

@ -0,0 +1,113 @@
package org.apache.commons.math3.genetics;
import java.util.List;
import org.apache.commons.math3.exception.DimensionMismatchException;
import org.apache.commons.math3.exception.MathIllegalArgumentException;
import org.apache.commons.math3.exception.NumberIsTooLargeException;
import org.junit.Assert;
import org.junit.Test;
public class NPointCrossoverTest {
@Test(expected = DimensionMismatchException.class)
public void testCrossoverDimensionMismatchException() {
final Integer[] p1 = new Integer[] {1,0,1,0,0,1,0,1,1};
final Integer[] p2 = new Integer[] {0,1,1,0,1};
final BinaryChromosome p1c = new DummyBinaryChromosome(p1);
final BinaryChromosome p2c = new DummyBinaryChromosome(p2);
final CrossoverPolicy cp = new NPointCrossover<Integer>(1);
cp.crossover(p1c,p2c);
}
@Test(expected = NumberIsTooLargeException.class)
public void testNumberIsTooLargeException() {
final Integer[] p1 = new Integer[] {1,0,1,0,0,1,0,1,1};
final Integer[] p2 = new Integer[] {0,1,1,0,1,0,1,1,1};
final BinaryChromosome p1c = new DummyBinaryChromosome(p1);
final BinaryChromosome p2c = new DummyBinaryChromosome(p2);
final CrossoverPolicy cp = new NPointCrossover<Integer>(15);
cp.crossover(p1c,p2c);
}
@Test(expected = MathIllegalArgumentException.class)
public void testCrossoverInvalidFixedLengthChromosomeFirst() {
final Integer[] p1 = new Integer[] {1,0,1,0,0,1,0,1,1};
final BinaryChromosome p1c = new DummyBinaryChromosome(p1);
final Chromosome p2c = new Chromosome() {
public double fitness() {
// Not important
return 0;
}
};
final CrossoverPolicy cp = new NPointCrossover<Integer>(1);
cp.crossover(p1c,p2c);
}
@Test(expected = MathIllegalArgumentException.class)
public void testCrossoverInvalidFixedLengthChromosomeSecond() {
final Integer[] p1 = new Integer[] {1,0,1,0,0,1,0,1,1};
final BinaryChromosome p2c = new DummyBinaryChromosome(p1);
final Chromosome p1c = new Chromosome() {
public double fitness() {
// Not important
return 0;
}
};
final CrossoverPolicy cp = new NPointCrossover<Integer>(1);
cp.crossover(p1c,p2c);
}
@Test
public void testCrossover() {
Integer[] p1 = new Integer[] {1,0,1,0,1,0,1,0,1};
Integer[] p2 = new Integer[] {0,1,0,1,0,1,0,1,0};
BinaryChromosome p1c = new DummyBinaryChromosome(p1);
BinaryChromosome p2c = new DummyBinaryChromosome(p2);
final int order = 3;
NPointCrossover<Integer> npc = new NPointCrossover<Integer>(order);
// the two parent chromosomes are different at each position, so it is easy to detect
// the number of crossovers that happened for each child
for (int i=0; i<20; i++) {
ChromosomePair pair = npc.crossover(p1c,p2c);
Integer[] c1 = new Integer[p1.length];
Integer[] c2 = new Integer[p2.length];
c1 = ((BinaryChromosome) pair.getFirst()).getRepresentation().toArray(c1);
c2 = ((BinaryChromosome) pair.getSecond()).getRepresentation().toArray(c2);
Assert.assertEquals(order, detectCrossoverPoints(p1c, p2c, (BinaryChromosome) pair.getFirst()));
Assert.assertEquals(order, detectCrossoverPoints(p2c, p1c, (BinaryChromosome) pair.getSecond()));
}
}
private int detectCrossoverPoints(BinaryChromosome p1, BinaryChromosome p2, BinaryChromosome c) {
int crossovers = 0;
final int length = p1.getLength();
final List<Integer> p1Rep = p1.getRepresentation();
final List<Integer> p2Rep = p2.getRepresentation();
final List<Integer> cRep = c.getRepresentation();
List<Integer> rep = p1Rep;
for (int i = 0; i < length; i++) {
if (rep.get(i) != cRep.get(i)) {
crossovers++;
rep = rep == p1Rep ? p2Rep : p1Rep;
}
}
return crossovers;
}
}